Orchid Historical Biogeography, Diversification, Antarctica and The

Journal of Biogeography (J. Biogeogr.) (2016) ORIGINAL Orchid historical biogeography, ARTICLE diversification, Antarctica and the paradox of orchid dispersal Thomas J. Givnish1*, Daniel Spalink1, Mercedes Ames1, Stephanie P. Lyon1, Steven J. Hunter1, Alejandro Zuluaga1,2, Alfonso Doucette1, Giovanny Giraldo Caro1, James McDaniel1, Mark A. Clements3, Mary T. K. Arroyo4, Lorena Endara5, Ricardo Kriebel1, Norris H. Williams5 and Kenneth M. Cameron1 1Department of Botany, University of ABSTRACT Wisconsin-Madison, Madison, WI 53706, Aim Orchidaceae is the most species-rich angiosperm family and has one of USA, 2Departamento de Biologıa, the broadest distributions. Until now, the lack of a well-resolved phylogeny has Universidad del Valle, Cali, Colombia, 3Centre for Australian National Biodiversity prevented analyses of orchid historical biogeography. In this study, we use such Research, Canberra, ACT 2601, Australia, a phylogeny to estimate the geographical spread of orchids, evaluate the impor- 4Institute of Ecology and Biodiversity, tance of different regions in their diversification and assess the role of long-dis- Facultad de Ciencias, Universidad de Chile, tance dispersal (LDD) in generating orchid diversity. 5 Santiago, Chile, Department of Biology, Location Global. University of Florida, Gainesville, FL 32611, USA Methods Analyses use a phylogeny including species representing all five orchid subfamilies and almost all tribes and subtribes, calibrated against 17 angiosperm fossils. We estimated historical biogeography and assessed the importance of different regions for rates of speciation, extinction and net species diversification. We evaluated the impact of particular LDD events on orchid diversity by asking how many species evolved in the new range subsequent to those events. Results Orchids appear to have arisen in Australia 112 Ma (95% higher prob- ability distribution: 102.0–120.0 Ma), then spread to the Neotropics via Antarc- tica by 90 Ma (HPD: 79.7–99.5 Ma), when all three continents were in close contact and apostasioids split from the ancestor of all other orchids. Ancestors of vanilloids, cypripedioids and orchidoids+epidendroids appear to have origi- nated in the Neotropics 84–64 Ma. Repeated long- and short-distance dispersal occurred through orchid history: stochastic mapping identified a mean total of À 74 LDD events or 0.8 Ma 1. Across orchid history, Southeast Asia was the most important source and maximally accelerated net diversification; across epidendroids, the Neotropics maximally accelerated diversification. Main conclusions Our analysis provides the first biogeographical history of the orchids, implicating Australia, the Neotropics and Antarctica in their origin. LDD and life in the Neotropics – especially the Andes – had profound effects on their spread and diversification; > 97% of all orchid species are restricted to individual continents. *Correspondence: Thomas J. Givnish, Department of Botany, University of Keywords Wisconsin-Madison, 430 Lincoln Drive, Asparagales, BioGeoBEARS, BiSSE, long-distance dispersal, Neotropics, Madison, WI 53706, USA. E-mail: [email protected] Southeast Asia vascular plants, grow in almost all terrestrial habitats except INTRODUCTION the driest deserts and are native to all continents except Orchids are the largest family of angiosperms, with roughly Antarctica (Pridgeon et al., 1999–2014; The Plant List, 2015). 880 genera and 27,800 species. They comprise ~8% of all Givnish et al. (2015) recently showed that the drivers of ª 2016 John Wiley & Sons Ltd http://wileyonlinelibrary.com/journal/jbi 1 doi:10.1111/jbi.12854 T. J. Givnish et al. extraordinary orchid diversity include the evolution of is that ability limited? Where did the key plant traits that pollinia, epiphytic habit, CAM photosynthesis, pollination appear to have accelerated net rates of orchid diversification via Lepidoptera, euglossine bees, and deceit, and life in arise? extensive tropical cordilleras such as the Andes and New Answers to these questions have been blocked by the lack Guinea Highlands. The defining characteristics of orchids – of a well-resolved, strongly supported backbone phylogeny minute seeds, germination aided by fungal symbionts, and for the orchids. Givnish et al. (2015) recently derived such a floral column of fused male and female parts – appear not phylogeny using a phylogenomics approach, calibrated to have accelerated orchid speciation themselves but to have against time using 17 angiosperm fossils. Here, we use this interacted in several cases with the traits just mentioned to tree to estimate the historical biogeography of Orchidaceae, generate exceptional levels of orchid diversity. identify its area of origin, assess the roles of vicariance and Across angiosperms, species richness is greater in families intercontinental dispersal, and locate the origin of epi- with broader geographical ranges and latitudinal extents phytism, a key trait that appears to have accelerated orchid (Ricklefs & Renner, 1994). Indeed, the area of ecozones or diversification (see also Chomicki et al., 2015). We test continents occupied explains 50% of the variance in log- whether occurrence on different continents had a significant transformed species richness in a phylogenetically structured effect on rates of speciation, extinction and net species diver- analysis of 409 angiosperm families (Vamosi & Vamosi, sification. Finally, we assess the extent by which long-dis- 2011). There remains the central question of whether large tance dispersal (LDD) may have increased orchid diversity areas or ecological volumes occupied cause large numbers of by asking how many species evolved in new ranges after species or vice versa (Ricklefs & Renner, 1994, 2000; Dodd LDD events. These across-family biogeographical analyses et al., 1999; Givnish et al., 2014). But given that orchids complement those that have been conducted on several smal- occur on all continents save Antarctica, and have one of the ler groups of orchids (Gravendeel et al., 2004; Bouetard widest latitudinal ranges of any plant family – from 72°N for et al., 2010; Smidt et al., 2011; Guo et al., 2012, 2015; Dueck Corallorhiza trifida in the Canadian Arctic Archipelago to et al., 2014; Freudenstein & Chase, 2015). 55°S for Chloraea, Codonorchis and Gavilea in Tierra del Fuego (Pridgeon et al., 1999–2014) – it seems plausible that MATERIALS AND METHODS high orchid diversity may partly reflect their broad distribution. Phylogenetics and tree calibration This raises what we term the ‘paradox of orchid dispersal’. On one hand, the broad distribution of orchids (and their Our study uses the supermatrix tree of Givnish et al. (2015), great diversity) might partly reflect their excellent dispersal, based on sequences of 75 plastid genes for 39 orchid species conferred by the dust-like seeds of almost all species. On the and 96 angiosperm outgroups, and three plastid genes for other hand, frequent long-distance seed dispersal should another 162 orchid species. This analysis includes placehold- work against differentiation within species and thus, ulti- ers for all five subfamilies, 18 of 19 tribes, and 40 of 43 sub- mately, speciation (Givnish, 2010). Dressler (1981) suggested tribes recognized by Chase et al. (2003), collectively that only 34 orchid genera had been able to cross tropical representing 99.6% of all orchid species. We continue using oceans and establish disjunct distributions on continents the tribal and subtribal nomenclature of Chase et al. (2003) long isolated from each other (i.e. South America, Africa and rather than its recent re-arrangement by Chase et al. (2015), Southeast Asia). Such disjunctions are seen within genera or to make our results regarding biogeography directly compa- pairs of closely related genera in all orchid subfamilies except rable to those in our earlier article devoted to plant traits Apostasioideae, suggesting that they may often have arisen (Givnish et al., 2015). early enough for continental drift to affect present-day distri- Givnish et al. (2015) calibrated the single maximum-likeli- butions (Dressler, 1981; Chase, 2001; but see Guo et al., hood tree resulting from the supermatrix analysis against the 2012). Such inferences are not based, however, on actual ages of 17 angiosperm fossils using beast 1.80 (Drummond calculations tied to the ages of specific fossils. Only three et al., 2012) and branch lengths based on atpB, psaB and orchid species reached the Hawaiian Islands by natural rbcL sequences. Dates and 95% highest posterior densities means (Wagner et al., 1990), suggesting limits imposed by (hereafter, HPD) were calculated for each node. Here, we seed dispersal or by missing pollinators or fungal symbionts. replace our sampling of Cypripedioideae with the more Dozens of closely related species of Teaguiea appear to have extensive set of taxa studied by Guo et al. (2012). We were speciated within a few kilometres of each other in Andean unable to incorporate these directly in our dating analyses, Ecuador (Jost, 2004), which also seems consistent with as the only locus shared between the two datasets was rbcL. short-distance dispersal of orchid seeds or their mutualists Instead, we replicated the BEAST analysis of Guo et al. and (Givnish et al., 2015). grafted the resulting time-calibrated phylogeny of cypri- Where did the orchids originate? To what extent do rela- pedioids onto ours. To refine modelling of geographical tionships above the generic level reflect intercontinental dis- diversification in tribe

Load more